Computationally Efficient Automatic Coast Mode Target Tracking Based on Occlusion Awareness in Infrared Images

Area of Science:

• Infrared imaging
• Computer vision
• Target tracking systems

Background:

• Centroid trackers are widely used in infrared imaging due to the difficulty of feature extraction.
• These trackers are vulnerable to target occlusion by clutter or background, leading to track loss.
• Coast mode tracking maintains tracking rates before occlusion but requires manual intervention.

Purpose of the Study:

• To propose an automatic coast mode tracking algorithm for centroid trackers in infrared images.
• To enhance target tracking robustness against occlusion and improve re-acquisition capabilities.
• To overcome limitations of traditional centroid trackers in challenging infrared environments.

Main Methods:

• Predicting target occlusion by analyzing brightness and potential screening objects.
• Implementing inertial tracking commands to maintain servo slew rate during predicted occlusion.
• Re-locking the target using histogram ratio-based target modeling after occlusion.

Main Results:

• The proposed algorithm demonstrates effectiveness in computer simulations with diverse infrared imagery sequences.
• Experimental results show improved tracking performance compared to existing published tracking algorithms.
• Validation in real-world environments using naval electro-optical tracking systems and airborne EO/IR systems.

Conclusions:

• The automatic coast mode tracking algorithm effectively mitigates target loss due to occlusion in infrared imaging.
• The method enhances the reliability of centroid trackers by enabling autonomous occlusion prediction and target re-acquisition.
• This approach offers a significant advancement for infrared tracking systems in complex operational scenarios.